lable at ScienceDirect

Placenta 40 (2016) 40e51

Contents lists avai

Placenta

journal homepage: www.elsevier .com/locate/placenta

What was the ancestral function of decidual stromal cells? A model forthe evolution of eutherian pregnancy

Arun Rajendra Chavan a, b, Bhart-Anjan S. Bhullar c, Günter P. Wagner a, b, d, e, *

a Department of Ecology and Evolutionary Biology, Yale University, New Haven, CT, 06511, USAb Yale Systems Biology Institute, Yale University, West Haven, CT, 06516, USAc Department of Geology and Geophysics, Yale University, New Haven, CT, 06511, USAd Department of Obstetrics, Gynecology and Reproductive Sciences, Yale Medical School, USAe Department of Obstetrics and Gynecology, Wayne State University, Detroit, MI, USA

a r t i c l e i n f o

Article history:Received 21 December 2015Received in revised form15 February 2016Accepted 21 February 2016

Keywords:Decidual stromal cellsEndometrial stromal cellsEutheriaExtended gestationOrigin of placental mammals

* Corresponding author. Department of Ecology anUniversity, New Haven, CT, 06511, USA.

E-mail address: gunter.wagner@yale.edu (G.P. Wag

http://dx.doi.org/10.1016/j.placenta.2016.02.0120143-4004/© 2016 Elsevier Ltd. All rights reserved.

a b s t r a c t

In human and mouse, decidual stromal cells (DSC) are necessary for the establishment (implantation)and the maintenance of pregnancy by preventing inflammation and the immune rejection of the semi-allograft conceptus. DSC originated along the stem lineage of eutherian mammals, coincidental with theorigin of invasive placentation. Surprisingly, in many eutherian lineages decidual cells are lost after theimplantation phase of pregnancy, making it unlikely that DSC are necessary for the maintenance ofpregnancy in these animals. In order to understand this variation, we review the literature on the fetal-maternal interface in all major eutherian clades Euarchontoglires, Laurasiatheria, Xenarthra and Afro-theria, as well as the literature about the ancestral eutherian species. We conclude that maintainingpregnancy may not be a shared derived function of DSC among all eutherian mammals. Rather, wepropose that DSC originated to manage the inflammatory reaction associated with invasive implantation.We envision that this happened in a stem eutherian that had invasive placenta but still a short gestation.We further propose that extended gestation evolved independently in the major eutherian cladesexplaining why the major lineages of eutherian mammals differ with respect to the mechanismsmaintaining pregnancy.

© 2016 Elsevier Ltd. All rights reserved.

1. Introduction reprogramming of gene regulatory status [8] and changes to the

Evolution of pregnancy is one of the salient attributes ofeutherian mammals. Distinctive features of eutherian pregnancyinclude invasive placentation (at least ancestrally), extendedgestation, maternal recognition of pregnancy, and the origin ofdecidual stromal cells (DSC).

DSCs are a novel cell-type that originated along the stem lineageof eutherian mammals [39,50]. In many eutherian mammals, DSCdifferentiate from a population of fibroblast-like cells in the uteruscalled Endometrial Stromal Fibroblasts (ESF) in a process calleddecidualization. In many eutherian mammals ESF decidualizeduring pregnancy, and in some eutherians, including humans, alsoduring the secretory phase of a sterile sexual cycle [29]. Human ESFdecidualize in response to progesterone and cyclic-AMP (cAMP).

Differentiation of DSC from ESF involves extensive

d Evolutionary Biology, Yale

ner).

genome-wide patterns of histone-modification [71]. They also ac-quire a distinct cellular morphologyeincreased size, globular orpolygonal appearance compared to the spindle-like appearance oftheir precursor fibroblasts, and increased accumulation of fat andglycogen granules as well as secretion of extracellular matrix [52].

1.1. Functions of decidual stromal cells

The functions of DSC have been studied extensively in modelsystems such as rodents and in-vitro grown human endometrialstromal cells. In human and mouse DSC form a physical barrier be-tween the invading syncytio-trophoblast and the maternal tissue.Failure to form this barrier in human leads to a pathological invasionof myometrium by the trophoblast, a condition called placentaaccreta that can be fatal to themother [30]. The presence of glycogenand lipid granules suggests a nutritive function toward the fetus [52].In addition DSC produce a variety of signaling molecules includingprolactin, prostaglandins, relaxin, IGFBP1 (Insulin-Like Growth Fac-tor Binding Protein 1) and many more. These hormones and

A.R. Chavan et al. / Placenta 40 (2016) 40e51 41

paracrine factors are important formaintainingmaternal physiologyin a state conducive to pregnancy [81]. In humans there is also evi-dence that thedeciduaplays a role in embryo selection, ensuring thatonly viable blastocysts can implant successfully [45].

Eutherian blastocyst implantation is an inflammatory process[19,77] and inflammation is necessary for successful implantation.Never the less soon after implantation the local endometrial envi-ronment becomes anti-inflammatory, a step necessary for themaintenance of pregnancy, mediated, in part, by a switch of thedecidual cell cytokine profile [65].

The fetus expresses paternal antigens that can be identified bymaternal immune system as ‘non-self’. Yet, the semi-allograft fetusis not rejected by the maternal immune system. How this fetal-maternal immune tolerance is brought about is a long-standingpuzzle in reproductive biology. In the last few years, DSC havebeen recognized as critical mediators of immunological tolerance atthe fetal-maternal interface in human and mouse [26].

In the mouse, through epigenetic silencing of the chemokines,Cxcl9 (C-X-C Motif Chemokine 9) and Cxcl10 (C-X-C Motif Chemo-kine 10), DSC limit the influx of cytotoxic T-cells into the endo-metrium, thus minimizing the interactions between thetrophoblast and effectors of the immune system [55]. Uterine var-iants of natural killer cells (uterine Natural Killer cells or uNK cells)and macrophages are distinct from their circulating counterparts inbeing less cytotoxic and active players in remodeling of uterinevasculature to accommodate invasive placentation. Acquisition oftheir distinct status in the endometrium is mediated byInterleukin-15 (IL15) secreted by decidual cells [26,40]. Condi-tioned medium from DSC, supplemented with IL15, converts pe-ripheral natural killer cells to their uterine phenotype [36], and co-culture of CD34-positive hematopoietic precursor cells with DSCconverts the precursor cells into uNK cells [74]. Decidualization inmouse creates an environment that prevents the growth oflymphatic vasculature in the endometrium, trapping dendritic cellsin the endometrium [15]. Dendritic cells are antigen-presentingcells that must traverse to a lymph node through lymphaticvasculature to present antigens to lymphocytes. This importantevent in the activation of the adaptive immune system is thusinterrupted by decidualization, at least in mouse.

Evidently, in human and mouse, DSC play a critical role inmodulation of the uterine environment to facilitate and maintainthe extended gestation in the face of immunological and physio-logical challenges of an invasive placenta.

1.2. Evolutionary origin of decidual stromal cells

DSC originated along the eutherian stem lineage as inferred byphylogenetic ancestral state reconstruction [50]. The evolution offunctional interactions between certain transcription factorsnecessary for DSC differentiation also occurred at the same time inphylogeny, supporting this inference. DSC differentiation is depen-dent on functional cooperative interactions between HOXA11 (Ho-meobox A11) and FOXO1 (Forkhead Box O1) [43], and betweenFOXO1 and CEBPB (CCAAT/Enhancer Binding Protein, Beta) [13].These cooperative interactions evolved along the stem lineage ofeutherian mammals: reconstructed ancestral eutherian versions ofthese transcription factors have the ability to up-regulate theexpression of DSC markers, while the reconstructed ancestraltherian versions lack this ability as do the proteins of outgroupspecies, opossum, platypus and chicken (HOXA11: [9]; CEBPB: [44]).

While mammalian viviparity and direct fetal-maternal contactand interaction likely evolved in the stem lineage of therian mam-mals (i.e. prior to the ancestor of both marsupial and eutherianmammals), the fetal-maternal interaction is qualitatively differentbetween metatherian (marsupial) and eutherian mammals. Highly

invasive forms of placentation that lead to a sustainable accommo-dation of the fetal allograft are unique to eutherian mammals [53].

Placentationhasbeencategorized into threemajor typesbasedonthe maternal tissue coming in direct contact with the trophoblast:epitheliochorial, endotheliochorial and haemochorial. Epi-theliochorial (trophoblast is incontactwith the luminal epitheliumofendometrium) placentation is non-invasive because the fetal tissuedoes not breach the uterine luminal epithelium [32]. It is found incattle, sheep, pig and horse and their relatives like dolphins andwhales. Endotheliochorial (trophoblast is in contact with maternalendothelium) and haemochorial (trophoblast is in contact withmaternal blood) are invasive forms because fetal tissue breaches theluminal epitheliumandestablishes adirect contactwithendometrialstroma. Endotheliochorial placentation is found in carnivores,elephant etc. and haemochorial placentation is found in primates,rodents, armadillo etc. [53]. Phylogenetic ancestral state re-constructionshave inferred that the eutherian ancestor possessed aninvasive form of placentation. Disagreement remainswhether it wasendotheliochorial [50] or haemochorial [22,79]; what is clear, how-ever, is that placentationwas invasive in the eutherian ancestor [47].In either case, the ancestor of eutherians had a placenta where thetrophoblast of the conceptus was in direct contact with the endo-metrial stroma. See Fig. 1 for a phylogeny of mammals and Fig. 2 forthe evolution of mode of placentation in Eutheria.

DSC are typically found in species that exhibit invasive placen-tation, with the possible exception of armadillo and other xenar-thrans, discussed below. After their origin in the eutherian stemlineage, DSC are reconstructed to have been lost in the lineageleading from the laurasiatherian ancestor, the same lineage inwhich invasive mode of placentation was lost [50].

When DSC originated is relatively clear, based on multiple linesof evidence mentioned above. However, which evolutionary forcesdrove their origin, what their ancestral function was, and whichontogenetic and gene regulatory changes made their originpossible are open questions. It is important to address thesequestions for at least two reasons. First, how and why evolutionarynovelties such as novel cell-types originate is a fundamentalquestion in evolutionary biology [4,76]. Secondly, understandinghow and why DSC originated can inform efforts to dissect themechanistic basis of reproductive pathologies.

In order to understand the origin and ancestral function of DSC,research on rodent and primate model systems needs to be sup-plemented with data on the other major lineages of eutherians, inparticular Afrotheria (tenrec, hyrax, elephant, manatee etc.) andXenarthra (armadillo, anteater, sloth etc.). Primates and rodents aremembers of one of the four major eutherian clades, Euarchonto-glires. Given that DSC originated in the eutherian stem lineage, it isimperative that any inferences concerning their origin be drawnfrom studies on taxa that bracket the entire diversity of eutheriandescent: Xenarthra, Afrotheria and Laurasiatheria, in addition toEuarchontoglires.

To this end, we reviewed the literature on DSC in Eutheria, withspecific attention to Xenarthra (armadillo) and Afrotheria (tenrecand hyrax). A surprising observation about DSC is that they aregenerally present in the peri-implantation phase, but tend todisappear in later stages of pregnancy. The latter observation isincompatible with the hypothesis that maintenance of extendedpregnancy is a shared derived eutherian function of DSC.

2. The life cycle of DSC and its implications for their ancestralfunction

2.1. DSC do not generally persist until the end of gestation

In species that have decidual cells, their numbers dwindle as

Fig. 1. Phylogeny of mammals. (a) Phylogenetic relationship between extant mammal species. Four major clades of Eutheria are coloured differently. Geological time scale ispresented on the right hand side. Names of the species discussed in detail in this article are in bold typeface. The tree was drawn based on branch-lengths from Ref. [21]. Therelationship between Xenarthra, Afrotherina and Boreoeutheria is not clearly resolved so far. Three plausible relationships are shown in (b) (c) and (d).

A.R. Chavan et al. / Placenta 40 (2016) 40e5142

Fig. 2. Evolution of mode of placentation according to Wildman and colleagues [79]. The tree presents the results of phylogenetic ancestral state reconstruction of mode ofplacentation. The tree topology used for this analysis by the authors is that in Fig. 1(c). Note that the reconstructed mode for the eutherian ancestral node is hemochorial.

A.R. Chavan et al. / Placenta 40 (2016) 40e51 43

pregnancy progresses. Mossman has pointed this out earlier in caseof human and rodents [52]. Here we call attention to this phe-nomenon in a select set of eutherian species representing Euarch-ontoglires, Xenarthra, Afrotheria and Laurasiatheria.

2.1.1. Euarchontoglires (Rat, Rattus norvegicus)In rat, similar to the situation in mouse, there are two sites of

decidualization in the uterus. Primary decidualization takes placeon the antimesometrial side in response to the implantation of theblastocyst. It is followed by secondary decidualization on themesometrial side [1]. Mesometrial decidua forms the deciduabasalis. Antimesometrial decidua encapsulates the conceptus, thusforming the decidua capsularis. The thickness of antimesometrialdecidua consistently decreases from day 8 of gestation before itcompletely disappears by day 18, 2e3 days prior to parturition [20].Similarly mesometrial decidua begins regression on day 14 andcontinues to regress until the end of gestation [18,28]. Severalstudies have shown that this regression is mediated by apoptoticcell death [18,33,66] (see Fig. 3a).

2.1.2. Xenarthra (Nine-banded armadillo, Dasypus novemcinctus)In the nine-banded armadillo (Dasypus novemcinctus), the

trophoblast comes in contact with the endometrial stromal cells atthe time of implantation. The question whether armadillo has DSCis still not completely settled. Enders and colleagues [23] describeda small number of cells that have the histological signs of DSC, butthere certainly is no compact layer of DSC comparable to the situ-ation in rodents and primates, and other authors deny the existenceof a decidua [7]. In the definitive (fully developed) placenta, how-ever, the trophoblast is in close contact with the myometrium; i.e.

there is no decidua basalis [23]. Our immunohistochemical inves-tigation of armadillo placenta corroborates this observation (Cha-van et al. in prep.). The only remnants of the endometrial stromaare thin bands separating the lobules of the placenta and a thinband encapsulating the placenta. Enders and colleagues [23] havecalled the latter ‘the endometrial arcade’; it is not a decidua cap-sularis because it does not surround the fetus, but only the placenta.This situation develops because the trophoblast does not invade theendometrium on a broad front, like in mouse or human, but sendsfinger like projections into pre-formed blood sinuses, which thenbranch to form a villous tree inside the maternal blood space [24].In essence, there is substantial amount of endometrial stromapresent at the time of implantation but all of it disappears from thebasal side of the definitive placenta, leaving the myometrium inclose proximity to the trophoblast (see Fig. 3b and c). Whether theendometrial arcade contains DSC is still unclear. A similar situationhas been described for another xenarthran species, the ant eaterTamandua tetradactylus [6]. Early in pregnancy a broad deciduali-zation of the stroma is found but at the site of placentation thedecidua are completely lost in later stages of gestation.

2.1.3. Afrotheria (Rock hyrax, Procavia capensis & Lesser hedgehogtenrec, Echinops telfairi)

According to Thursby-Pelham's description of the uterus of hy-rax, Procavia capensis, the decidual layer is at its thickest in theearliest specimen studied (specimen 3, uterine diameter 8 mm),which gradually thins out in specimens at later stages (specimen 4,uterine diameter 15 mm and specimen 5, uterine diameter 17 mm)[72]. Our examination of histological preparations of hyrax(P. capensis) uterus (ID 20487 from Mossman Developmental

Fig. 3. Decidual stromal cells decrease over the length of gestation. (a) Rat: thickness of decidual layer is plotted against day of gestation. Thickness of all three layers of deciduadecreases over time. Data for anti-mesometrial and lateral decidual thickness was obtained from Ref. [20] and data for mesometrial decidual thickness was obtained from Ref. [28].(b) (c) Armadillo: histological preparations of armadillo fetal-maternal interface stained for cytokeratin (brown). Cytokeratin marks trophoblast cells and glandular and luminalepithelia of the endometrium. In the early stage, the peri-implantation phase, endometrial stroma is present, and the stromal cells have DSC morphology. In the later stage, note theabsence of endometrial layer between the placental villi and myometrium. (d) (e) Hyrax: Haematoxylin and Eosin stained slides of fetal-maternal interface of hyrax from MossmanCollection. Thickness of decidual layer in the early stage uterus (uterine diameter 8 mm) is 1850m, which reduces to 750m in a later stage uterus (uterine diameter 13 mm). (f) (g)Tenrec: fetal-maternal interface from an early stage of gestation (crown-rump length 5e8 mm), before the formation of definitive placenta, with permission, from Ref. [11] and alater stage of gestation (crown-rump length 58 mm), after the formation of definitive placentation, with permission, from Ref. [10]. Note that the endometrial stroma is present inthe early stage, but completely disappears in the later stage. These images have been reproduced with permission from the publisher, and cropped to show the relevant parts of theimages. Myo ¼ myometrium, Tr ¼ Trophoblast, Endo.S. ¼ Endometrial stroma. Scale-bars on (b), (c) and (f) are 100 m, and (g) is magnified 12�.

A.R. Chavan et al. / Placenta 40 (2016) 40e5144

Collection, Zoological Museum, University of Wisconsin, Madison)confirms these results (see Fig. 3d and e).

Carter and colleagues have described the definitive placenta of

tenrec, Echinops telfairi, and its development [10,11]. During thedevelopment of the placenta, stromal cells can be seen in theendometrium, which are potentially DSC. Carter and colleagues do

A.R. Chavan et al. / Placenta 40 (2016) 40e51 45

not call these cells ‘decidual’ based on their histological appearance,but further investigation is needed to resolve their identity. Thepossibility is open that these cells are differentiated decidualstromal cells (by their gene regulatory identity) but histologicallyinconspicuous. Despite their equivocal identity as ESF or DSC,endometrial stromal cells are lost entirely in the later stages ofpregnancy, eliminating the possibility of persistence of any DSCthrough gestation (see Fig. 3f and g).

2.1.4. Laurasiatheria (European mole, Talpa europaea; Indian falsevampire bat, Megaderma lyra lyra; and Neotropical disc-wingedbat, Thyroptera tricolor spix)

Laurasiatherian clades, Carnivora (e.g. cat, mink), Chiroptera(bats), and Eulipotyphla (e.g. European mole), have members withinvasive placentation. Carnivores are termed as ‘deciduate’ speciesbecause maternal tissue, mostly glandular epithelium and endo-thelium, is shed at birth. However, their ESF do not seem to undergodecidualization based on histological criteria [53]. The literature onDSC of Eulipotyphla and Chiroptera suggests that DSC are presentaround the peri-implantation period, but undergo degeneration inlater stages; sometimes all of them, e.g. in case of the Indian falsevampire bat, Megaderma lyra lyra, bringing the placental villi andmyometrium in apposition (Indian false vampire bat: [31];Neotropical disc-winged bat: [80]; European mole: [46]).

It is clear that in these species, drawn from all major clades ofEutheria, DSC are present, with possible exceptions, and the layer ofdecidual cells is at its thickest at the time of implantation. DSC tendto disappear in the later stages of pregnancy. This is particularlystriking in armadillo and tenrec, in which decidua basalis seems tobe completely abolished once the definitive placenta is established- a situation that can be called a ‘physiological placenta accreta’ e acondition clearly pathological in human but normal in theseanimals.

3. The ancestral function of DSC is its role duringimplantation

The taxonomic distribution of the life cycle of DSC describedabove suggests that the presence of DSC is a shared derived state ofeutherian mammals. Furthermore the evidence suggests that laterstages of pregnancy are not homologous [70]. This leads to theconclusion that, ancestrally, DSC had a role to play during earlystages of pregnancy. Their endocrine and immune functions in laterstages of pregnancy seem to be limited to Euarchontoglires ratherthan being general to Eutheria. When this observation is placed inthe context of what we know about mammalian evolution, thefollowing model emerges. Our model is consistent with a similarargument proposed by Swaggart and colleagues, namely that thelater stages in the pregnancy of eutherian taxa are not homologous[70].

3.1. Model of evolution of extended gestation and ancestral functionof DSC

We propose that, in the eutherian stem lineage, gestation wasshorter than or at most as long as the sterile sexual cycle, i.e. theancestral eutherian did not have ‘extended gestation’ similar to thesituation in basal marsupial lineages, e.g. the opossums [62].

To our knowledge, a term for length of gestation in relation tothe length of the sterile sexual cycle doesn't exist. Gestation longerthan sterile sexual cycle is often referred to as ‘extended gestation’,but this term can also be used to describe the absolute length ofgestation, not relative to that of the sterile sexual cycle. We wouldlike to propose the terms ‘intra-cyclic gestation’ and ‘trans-cyclicgestation’ to refer to gestation shorter and longer than the sterile

sexual cycle, respectively. We follow this terminology throughoutthis article from here on. See Fig. 4 for an illustration explainingintra-cyclic and trans-cyclic gestations.

We propose that trans-cyclic gestation originated indepen-dently in the four major eutherian lineages. This proposal canexplain the wide range of differences in the role of DSC reviewedabove as well as other differences briefly summarized below. It isbased on the fact that species in basal marsupial lineages have shortintra-cyclic gestation, that the members of the eutherian stemlineage and probably also the eutherian ancestor were small ani-mals, and that the four major lineages of eutherians arose in a veryshort period of intense radiation (evidence reviewed below).

This scenario implies that when DSC originated in the stemeutherian lineage, gestation likely involved only a short post-implantation phase. The ancestral function of DSC was, therefore,limited to the implantation process, perhaps to dampen the in-flammatory response elicited by invasive implantation [51]. Ifeutherian pregnancy was extended beyond the length of sterilesexual cycle independently in major eutherian lineages, the endo-crine and immune mechanisms for maintaining trans-cyclicgestation were also acquired independently in each of the majorlineages and thus differ in the way pregnancy is maintained. Fromthis model we infer that the role of DSC in maintaining pregnancyin primates and rodents evolved in the stem lineage of theEuarchontoglires rather than in the stem lineage of Eutheria.

4. The paleontological and comparative physiologicalevidence

The validity of the model explained above hinges upon theassumption that the eutherian ancestor had an intra-cyclic gesta-tion. In this section we present evidence that substantiates thisassumption.

4.1. Trans-cyclic gestation probably originated independently inmajor eutherian lineages

Marsupial pregnancy is intra-cyclic [49]. Maternal hormonalcycle is not altered by the presence of a fetus; there is no so-called‘maternal recognition of pregnancy’ except in highly derived mac-ropodids, e.g. wallabies [62]. The condition in macropodids is not ashared derived character of marsupials [27].

The situation in basal marsupials is in sharp contrast to that ineutherians. Gestation going beyond the length of sterile sexualcycle, trans-cyclic gestation, is considered to be a hallmark ofeutherian pregnancy. There is maternal recognition of pregnancy:maternal physiology is altered by the presence of the fetus; inparticular, progesterone production is sustained through gestation[5].

The evidence detailed below supports our argument that trans-cyclic gestation had multiple independent origins.

4.1.1. The eutherian ancestor had a short gestationThe fossil record suggests that the eutherian ancestor was a very

small animal, and that body size increased independently in severallineages [57]. The high end of reconstructed ancestral sizes (in the100e200 g range) is likely biased by the secondary body size en-largements strongly suggested to have taken place along themonotreme and marsupial stems. In fact, fossil stem monotremesand stem metatherians are very small, as are stem eutherians [37].These would be closer to the 1e15 g range of small extant in-sectivores. We think that this is a conservative estimate since re-mains of larger animals are more likely to be recovered than thoseof very small animals.

Moreover, there is direct evidence that body sizes increased

Fig. 4. Intra-cyclic and trans-cyclic gestation. Intra-cyclic gestation is shorter than the sterile sexual cycle. Trans-cyclic gestation is longer than the sterile sexual cycle. Trans-cyclicgestation can be interpreted to have a new phase of gestation intercalated between implantation and parturition.

A.R. Chavan et al. / Placenta 40 (2016) 40e5146

along each of the major placental lineagesenot just on the stems ofthe four “main” clades but even on the various stem lineages ofsmaller “order level” clades within these. This pattern of body sizeincrease was noticed long ago and given the term “Cope's Rule”[34]; however, the concept is vague and not well controlledphylogenetically. The pattern of evolution along the stems of majorplacental lineages is further obfuscated by the fact that very few ofthe thousands of known Paleogene mammal taxa have beenrigorously placed phylogenetically. Much additional work remainsto be done. Nevertheless, along the earliest parts of the stem line-ages of the various placental clades are small-bodied animals. Theafrotherian and xenarthran records are unfortunately poor in theearly Paleogene. Among Euarchontoglires, rodents have retainedsmall body size for the most part, as have tree shrews and der-mopterans, successive sisters to the larger-bodied Primates. For themost part, Eulipotyphla and Chiroptera, “insectivorans” and bats,remain small-bodied and their fossil stem taxa are small-bodied aswell [67]. Early possible stem perissodactyls (“condylarths”), earlystem pangolins, i.e. scaly anteaters (“palaeanodonts”), and earlystem carnivorans (for instance from the Eocene of Germany) areshrew or mouse size [64]. The stem artiodactyl record is not ascomplete, but the earliest stem artiodactyls, while not as tiny as thestem members of other lineages, are roughly the size of large rab-bits [16,63]. A general pattern thus emerges of repeated instances ofextreme body enlargement.

Life history studies have consistently found a positive relation-ship between body weight (which scales with body size) and thelength of gestation [38,48]. With the use of phylogenetic compar-ative methods controlling for the effects of shared history, thisrelationship is less steep than was previously believed. The allo-metric scaling exponent is 0.18e0.2 by ordinary least squaresmethod but 0.07 to 0.1 by phylogenetic generalized least squaresmethod [14]. However, there is little reason to doubt the general

relationship between body size and gestation length outside oflower taxonomic levels, given that most time during long gesta-tions is dedicated to fetal growth, and thus has to relate to neonatalsize.

Given the small body sizes of the eutherian ancestor and theancestors of the major eutherian clades, and the relationship be-tween body size and gestation length, we infer that these animalslikely had short gestations. Whether this amounts to intra-cyclicgestation is not directly testable, but at least large body size inthe eutherian ancestor would be incompatible with the hypothesisof intra-cyclic gestation.

The idea that the eutherian ancestor gave birth to altricial ne-onates (typically defined by lack of hair and closed eyes at birth andthe need for extensive parental care) has existed for a long time[59]. This idea has been consistently supported by several studiesusing phylogenetic ancestral state reconstruction [50,57]. Polari-zation of the character is provided in part by the highly altricialnature of monotreme hatchlings, which are similar in manyways tomarsupial neonates. The fact that altricial neonates are born after arelatively short gestation, and depend largely on lactation for theirgrowth, is also suggestive of a short gestation in the eutherianancestor.

Admittedly, this reasoning alone doesn't lead us to the inferencethat the eutherian ancestor had an intra-cyclic gestation (i.e.gestation shorter than the sterile sexual cycle). However, we mustreflect upon the inferred short gestation of the eutherian ancestorin the context of the ancestral therian mode of reproduction, whichmost likely did not involve trans-cyclic gestation. More likely thanthe origin of trans-cyclic gestation in the eutherian stem lineage isthe possibility that characters such as invasive placentation andDSC originated first, paving the way for the evolution of muchlonger trans-cyclic gestation. Once the problem of inflammationcaused by implantation was solved, in part, through the anti-

A.R. Chavan et al. / Placenta 40 (2016) 40e51 47

inflammatory role of DSC and progesterone, further extension ofgestation was possible leading to trans-cyclic gestation.

4.1.2. Mechanisms involved in maintenance of trans-cyclicgestation are highly variable

Corpus luteum, the remnant of an ovarian follicle upon ovulation,secretes progesterone. In eutherian mammals, a successful preg-nancy requires sustained progesterone production, well beyond thelife span of the corpus luteum of a sterile sexual cycle. This can beachieved in twoways: by extension of the life span of corpus luteumor by another organ taking over the responsibility of progesteronesecretion after the corpus luteum has regressed. All eutherian taxause either one of these two mechanisms or both to maintain theirpregnancy beyond the length of a sterile sexual cycle. However, themeans by which they activate these mechanisms are remarkablydifferent [5]. See Fig. 5 for a summary of this section.

Corpus luteum can be rescued from regression either by pro-motion of its growth through luteotropic signals or by inhibition ofluteolysis through anti-luteolytic signals. In a sterile sexual cycle,prostaglandins, most often from the uterus, lead to luteolysis. Anti-luteolytic signals act to negate the effects of prostaglandins.

In Euarchontoglires, typically the life span of corpus luteum isextended by luteotropic signals, which may or may not be followedby a luteo-placental shift in progesterone production. The luteo-tropic signal in human, human Chorionic Gonadotrophin (hCG)from the trophoblast, promotes growth of corpus luteum throughpart of gestation, after which the placenta takes over progesterone

Fig. 5. Divergent mechanisms maintain trans-cyclic gestation in eutherian mammals. Mecbeyond the length of sterile sexual cycle) are grouped into four categories: luteotropic sigmechanisms. Note the remarkable diversity of the mechanisms. Roughly, EuarchontogliresaCL ¼ accessory corpora lutea, PRL ¼ prolactin, hCG ¼ human chorionic gonadotropinP4 ¼ progesterone.

production. The luteotropic signal in rodents is prolactin, likelysecreted by the decidua or the pituitary. It promotes growth ofcorpus luteum through the entire length of gestation, without aluteo-placental shift. Rabbits also do not have a luteo-placentalshift. They make use of an unidentified luteotropic signal ofplacental origin in addition to estradiol (E2) as the anti-luteolyticsignal [5].

In Laurasiatheria, the mechanisms are more variable than inEuarchontoglires. Artiodactyls typically use anti-luteolytic signals,along with or without a shift in the organ of progesterone pro-duction. In pecoran ruminants (sheep, cattle, goat) IFN- t secretedby the conceptus extends the life span of corpus luteum by sup-pressing endometrial prostaglandin-mediated luteolysis. Pigs alsouse an anti-luteolytic signal but it is E2 secreted by the pigtrophoblast, which converts the uterine prostaglandin secretionfrom endocrine to exocrine. This prevents the entry of prosta-glandin into the bloodstream, and therefore it cannot reach thecorpus luteum [5].

Horse (a perissodactyl) uses equine Chorionic Gonadotropin(eCG) as a luteotropic signal, which probably induces production ofaccessory corpora lutea. Accessory corpora lutea are ovarian folliclesthat luteinize without ovulation and contribute to progesteroneproduction. The placenta primarily carries out progesterone pro-duction in late gestation after regression of primary and accessorycorpora lutea [3].

Dog (a carnivore) possibly does not have a signal from theconceptus for extension of the life span of the corpus luteum, given

hanisms of maintenance of trans-cyclic gestation (sustained progesterone productionnals, anti-luteolytic signals, shift in the organ of progesterone production, and otheruse luteotropic signals and artiodactyls use anti-luteolytic signals. CL ¼ corpus luteum,, eCG ¼ equine chorionic gonadotropin, E2 ¼ estradiol, IFN- t ¼ interferon-tau,

A.R. Chavan et al. / Placenta 40 (2016) 40e5148

that the life span of corpus luteum is not altered during pregnancyin comparison to pseudo-pregnancy [5].

In armadillo (a xenarthran), the corpus luteum regresses duringblastocyst dormancy. After implantation progesterone productionis taken over by the adrenals of the fetus [54], possibly supple-mented by the placenta [41].

Elephant (an afrotherian) shows signs of neither a luteo-placental shift nor a luteotropic or anti-luteolytic signal. It de-velops accessory corpora lutea that secrete progesterone throughthe length of gestation [42,69].

In sum, sustained production of progesterone, past the life spanof corpus luteum of a sterile sexual cycle, is the pivotal modificationof maternal physiology during pregnancy that makes trans-cyclicgestation possible. Highly divergent mechanisms involved inbringing about this outcome strongly suggest the possibility thattrans-cyclic gestation is not an ancestral character of Eutheria, butthat it originated independently in various lineages, as also pro-posed by Swaggart and colleagues [70].

4.1.3. Epipubic bones in stem EutheriaEpipubic bones are a pair of bones projecting anterio-ventrally

from the pubis in monotremes and marsupials. They wereinitially thought to support the marsupium, i.e. the pouch in mar-supials and echidnas, but this view is contended given their pres-ence in species that do not have a pouch. They may indeed berelated to the presence of an indented pouch ‘region’ if not a fullpouch, but they have also been implicated in gait and locomotion.Specifically, the epipubic bones extend into the superficial layers ofhypaxial abdominal musculature, bearing a series of attachmentsand connections to the body wall, the midline, and the femur, andcontribute to a “cross-couplet” lever system during locomotion thatis more similar to the reptilian condition than the unilateral muscleactivation pattern of eutherian mammals [60,61,75].

Despite their uncertain function, it is likely that epipubic bonespreclude trans-cyclic gestation by compromising pliability of theabdominal wall, which is essential for accommodation of a growingfetus for an extended duration. Extant eutherians lack epipubicbones. Their loss was probably one of the factors that made trans-cyclic gestation possible or their loss was driven by the need toaccommodate numerous large-sized fetuses. Fossils of stem eu-therians, however, do possess the epipubic bones, as do stemmetatherians and stem therianseunambiguously indicating theirhomology across all these clades [56,75]. This indicates at least thatthese elements persisted until very near to, and perhaps into, theextant eutherian radiation, adding support for the argument thatancestral eutherian likely gave birth to small neonates and havehad intra-cyclic gestation.

Taken together, these data suggest that, despite its universalityin Eutheria, trans-cyclic gestation evolved independently severaltimes after the eutherian radiation that gave rise to the majorclades of placental mammals.

5. Integration of our model in the narrative of mammalianevolution

Below, we insert the ideas postulated above into the account ofmammalian evolution, as currently understood (see Fig. 6).

The eutherian ancestor was a very small mammal and gave birthto altricial neonates after a short gestation period. Gestation wasintra-cyclic, and contained within the luteal phase of the cycle, as itis in opossum and many other marsupials. The embryo remainedunattached for most of gestation, and implantation took place onlytoward the end of the pregnancy. This is similar to the scenario inextant marsupials such as opossums, and it is likely to have beenthe case in the marsupial ancestor as well as the therian ancestor.

The key difference to the situation in marsupials was that, in theeutherian ancestor, implantation was invasive. This suggests thatinvasive placentation, following the destruction of endometrialluminal epithelium, elicited an inflammatory response in theendometrium. Tissue damage is known to lead to the activation oftissue resident fibroblasts, which participate in the inflammatoryreaction similarly to tissue macrophages [68] and leads to theactivation of FOXO1 protein [2], a transcription factor known to becritical for decidualization [29]. We thus suggest that DSC origi-nated from ESF by converting the pro-inflammatory activated fi-broblasts of the endometrial stroma into a specialized stromal celltype that participates in the inflammatory implantation process butchanges the nature of the response to an anti-inflammatory statethat is compatible with accommodating the conceptus. Some of themechanisms of the switch from pro-to anti-inflammatory statehave been identified; for instance the bi-phasic expression of IL-33(Interleukin 33) related signaling molecules in human endometrialstromal fibroblasts during decidualization [65].

In our model the eutherian ancestor had intra-cyclic gestation,parturition ensued soon after a short phase of invasive placenta-tion. In this scenario, the role of DSC was limited to the peri-implantation period, possibly to modulate the inflammatory stim-ulus emanating from the blastocyst to limit tissue damage to theuterus.

Extinction of large terrestrial dinosaurs (excepting the small-bodied ancestors of the major avian lineages) at the Crea-taceousePaleogene boundary (K-Pg), 65 million years ago, mayhave opened up ecological niches for mammals, leading to amammalian radiation and several independent trends of increasein body size. Ancestors of the four major clades of eutherianmammals already existed before this mass extinction event [21];but see Ref. [57] for an alternate hypothesis); therefore eutherianbody size increased independently in several lineages after the K-Pgboundary.

Consequently, we hypothesize, that long, trans-cyclic gestationalso evolved in concert with and to accommodate the increase inbody size, independently in major eutherian lineages. Independentevolution of trans-cyclic gestation in turn makes it likely thatdifferent functional strategies evolved to sustain gestation. Trans-cyclic gestation can be thought of as the intercalation of a newphase of pregnancy between implantation and parturition (seeFig. 4) in the ancestral mode of pregnancy hypothesized above.Addition of this phase necessitated resolution of immunologicaland physiological challenges of a long pregnancy that is associatedwith an invasive placenta. According to our model these challengesarose independently in various eutherian lineages, and so did themechanisms of their resolutions. The resolution, in Euarchonto-glires alone, involved extended life span of DSC, together with theacquisition of new functions related to maintenance of the trans-cyclic gestation. It is not fully understood how these challengeswere resolved in Xenarthra and Afrotheria in a manner that doesnot involve DSC. It is possible that in Laurasiatheria, at least, theimmunological challenge was tackled in part by loss of invasiveplacentation [12].

6. Alternative models explaining reproductive variation

There is tremendous variation in the morphology and physi-ology of the fetal-maternal interface in therian mammals. Placentais the most variable organ, at least in Theria, despite the fact that itserves the same basic function in all lineages. A number of expla-nations have been put forth to explain this variation.

Parent-offspring conflict is often invoked as a driver of diversityin placental structures [17]. In this model, placental diversity is theresult of a constant tug-of-war between the mother and the fetus,

Fig. 6. Model for the evolution of eutherian gestation. Horizontal bars on the branches of the tree indicate evolutionary changes. Note the multiple origins of trans-cyclic gestationin Eutheria. The grey box indicates whether implantation and maintenance of trans-cyclic gestation are dependent on or independent of DSC. Intra-clade variation clearly exists, buthere we have tried to show the most prominent patterns.

A.R. Chavan et al. / Placenta 40 (2016) 40e51 49

particularly over characters related to resource transfer and thetiming of parturition. While this argument may apply to the di-versity within narrower clades of eutherians, we argue that thebroad pattern of diversity across eutherians does not need a se-lective explanation, because the trans-cyclic mode of pregnancy islikely not homologous across major eutherian clades. Convergentorigin of trans-cyclical gestation by different mechanisms is suffi-cient to explain diversity at this level of comparison.

Another possible explanation of mechanistic diversification isdevelopmental systems drift i.e. variation in the developmentalmechanisms of homologous characters [73,78]. One model fordevelopmental system drift is the Selection-Pleiotropy-Compensation (SPC) model [35,58]. The SPC model assumes thatselection of an adaptive mutation brings with it negative pleio-tropic consequences, which are subsequently resolved bycompensatory mutations. SPC model has been proposed as anexplanation for many puzzling phenomena including develop-mental systems drift. According to SPC model, development ofhomologous characters, when evolving under selection forcompensating mutations, can turn out to be surprisingly variableamong species. It can be argued that eutherian pregnancy is asvariable as it is because it has been undergoing developmentalsystems drift under the SPC mode of evolution. The model pro-posed here provides an alternative explanation. The SPC modelaims at explaining why homologous traits can be produced bydifferent mechanisms. Again, the alternative is that the trait, trans-cyclical gestation, is not homologous across the major clades ofeutherian mammals and thus it is not surprising that it is producedby different mechanisms.

We suspect that all three of these explanations; parent-offspringconflict, developmental systems drift and independent derivation;

are at work in producing the variation among modes of eutheriangestation, but at this point it is not clear at what taxonomic levelsthese different mechanisms act. Parent-offspring conflict and SPCperhaps explain the variation in the shared derived phase ofgestation i.e. the ancestral short post-implantation phase ofgestation (see Fig. 4) as well as variation found among more closelyrelated animals. Variation in the later stages of pregnancy that wereindependently acquired in various lineages can be explained byindependent origination. Swaggart and colleagues [70] point outthat during embryonic development, organogenesis is generallycompleted by the time of luteolysis. For example, in mouse, CLpersist till the end of gestation and the fetuses are altricial at thetime of birth, while in human luteolysis takes place around 8e12weeks of gestation coinciding with the completion of organogen-esis (i.e. the fetus is roughly in the stage of development that mousefetuses are at the time of birth) which is followed by a long phase ofintrauterine growth and maturation. Swaggart and colleaguessuggest that mouse gestation is homologous to only the firsttrimester of human gestation. The variation between mouse (andpotentially other altricial rodent species) gestation and the firsttrimester of human gestation is probably better explained by SPCand parent-offspring conflict than by our model, while the varia-tion between later stages of gestation of precocial mammals likeguinea pig, and second and third trimesters of human gestation areprobably better explained by the assumption of independent ori-gins. The evaluation of these ideas requires more detailed analysisof gestational variation than is the scope of this paper.

7. Conclusion

In light of the their critical immune and endocrine functions to

A.R. Chavan et al. / Placenta 40 (2016) 40e5150

support pregnancy in rodents and primates, it is surprising thatDSC in xenarthrans, afrotherians and some laurasiatherians, whilepresent during implantation, are lost in later stages of pregnancy.Here we explained this pattern by proposing that DSC in theeutherian ancestor played a role only in the peri-implantationperiod, perhaps to modify the inflammatory reaction elicited bythe process of invasive implantation. Their role in later gestation isnot shared among the major eutherian lineages, and probably wasacquired in the stem lineage of Euarchontoglires.

An obvious question arising from the model proposed here ishow the fetal-maternal immune tolerance is mediated in Xenarthraand Afrotheria as for the majority of their gestation proceedswithout any aid from DSC. Our understanding of this phenomenonis limited at the moment. One model, put forward by Enders andWelsh [25], suggests that one of the strategies is to limit the contactbetween the endometrial stroma and the trophoblast. More work iscertainly required on this front.

Finally, an implication of our model is that research on theevolutionary origin of DSC will benefit most from understandingthe biology of implantation and the role of DSC in that phase ofgestation rather than their role in the maintenance of trans-cyclicgestation.

Acknowledgments

We would like to thank the Zoological Museum at University ofWisconsin-Madison for lending to us histological preparationsfrom the Mossman Developmental Collection. Research into theevolution of the decidual cell type is funded by the John TempletonFoundation grant number 54860 “How new cell types arise inevolution”. Other research in the lab is funded by NSF grant IRES#14e001273, as well as the Yale University Science DevelopmentFund.

References

[1] P.A. Abrahamsohn, T.M. Zorn, Implantation and decidualization in rodents,J. Exp. Zool. 266 (6) (1993) 603e628.

[2] M. Alikhani, Z. Alikhani, D.T. Graves, FOXO1 functions as a master switch thatregulates gene expression necessary for tumor necrosis factor-inducedfibroblast apoptosis, J. Biol. Chem. 280 (13) (2005) 12096e12102.

[3] W. Allen, Hormonal control of early pregnancy in the mare, Anim. Reprod. Sci.7 (1) (1984) 283e304.

[4] D. Arendt, The evolution of cell types in animals: emerging principles frommolecular studies, Nat. Rev. Genet. 9 (11) (2008) 868e882.

[5] F.W. Bazer, T.E. Spencer, Chapter 5-hormones and pregnancy in eutherianmammals, in: D.O.N.H. Lopez (Ed.), Hormones and Reproduction of Verte-brates, Academic Press, London, 2011, pp. 73e94.

[6] H. Becher, Placenta und Uterusschleimhaut von Tamandua tetradactyla(Myrmecophaga), Gegenbaurs Morphol. Jahrb. 67 (1931) 381e458.

[7] K. Benirschke, Comparative Placentation, 2011. Retrieved December 13, 2015,from, http://placentation.ucsd.edu/homefs.html.

[8] A.K. Brar, S. Handwerger, C.A. Kessler, B.J. Aronow, Gene induction and cate-gorical reprogramming during in vitro human endometrial fibroblastdecidualization, Physiol. Genomics 7 (2) (2001) 135e148.

[9] K.J. Brayer, V.J. Lynch, G.P. Wagner, Evolution of a derived protein-proteininteraction between HoxA11 and Foxo1a in mammals caused by changes inintramolecular regulation, Proc. Natl. Acad. Sci. U. S. A. 108 (32) (2011)E414eE420.

[10] A.M. Carter, T.N. Blankenship, H. Kunzle, A.C. Enders, Structure of the defini-tive placenta of the tenrec, Echinops telfairi, Placenta 25 (2e3) (2004)218e232.

[11] A.M. Carter, T.N. Blankenship, H. Kunzle, A.C. Enders, Development of thehaemophagous region and labyrinth of the placenta of the tenrec, Echinopstelfairi, Placenta 26 (2e3) (2005) 251e261.

[12] A.M. Carter, A.C. Enders, The evolution of epitheliochorial placentation, Annu.Rev. Anim. Biosci. 1 (1) (2013) 443e467.

[13] M. Christian, X. Zhang, T. Schneider-Merck, T.G. Unterman, B. Gellersen,J.O. White, J.J. Brosens, Cyclic AMP-induced forkhead transcription factor,FKHR, cooperates with CCAAT/enhancer-binding protein beta in differenti-ating human endometrial stromal cells, J. Biol. Chem. 277 (23) (2002)20825e20832.

[14] M. Clauss, M.T. Dittmann, D.W.H. Muller, P. Zerbe, D. Codron, Low scaling of alife history variable: analysing eutherian gestation periods with and without

phylogeny-informed statistics, Mamm. Biol. 79 (1) (2014) 9e16.[15] M.K. Collins, C.S. Tay, A. Erlebacher, Dendritic cell entrapment within the

pregnant uterus inhibits immune surveillance of the maternal/fetal interfacein mice, J. Clin. Investig. 119 (7) (2009) 2062e2073.

[16] L.N. Cooper, J. Thewissen, S. Bajpai, B. Tiwari, Postcranial morphology andlocomotion of the Eocene raoellid Indohyus (Artiodactyla: Mammalia), Hist.Biol. 24 (3) (2012) 279e310.

[17] B. Crespi, C. Semeniuk, Parent-offspring conflict in the evolution of vertebratereproductive mode, Am. Nat. 163 (5) (2004) 635e653.

[18] D. Dai, B.C. Moulton, T.F. Ogle, Regression of the decidualized mesometriumand decidual cell apoptosis are associated with a shift in expression of Bcl2family members, Biol. Reprod. 63 (1) (2000) 188e195.

[19] N. Dekel, Y. Gnainsky, I. Granot, K. Racicot, G. Mor, The role of inflammationfor a successful implantation, Am. J. Reprod. Immunol. 72 (2) (2014) 141e147.

[20] A.D. Dickson, “The disappearance of the decidua capsularis and Reichert'smembrane in the mouse.”, J. Anat. 129 (Pt 3) (1979) 571e577.

[21] M.J. dos Reis Inoue, M. Hasegawa, R.J. Asher, P.C. Donoghue, Z. Yang, Phylo-genomic datasets provide both precision and accuracy in estimating thetimescale of placental mammal phylogeny, Proc. Biol. Sci. 279 (1742) (2012)3491e3500.

[22] M.G. Elliot, B.J. Crespi, Phylogenetic evidence for early hemochorial placen-tation in eutheria, Placenta 30 (11) (2009) 949e967.

[23] A.C. Enders, G.D. Buchanan, R.V. Talmage, Histological and histochemical ob-servations on the armadillo uterus during the delayed and postimplantationperiods, Anat. Rec. 130 (4) (1958) 639e657.

[24] A.C. Enders, A.M. Carter, Review: the evolving placenta: different develop-mental paths to a hemochorial relationship, Placenta 33 (Supplement) (2012)S92eS98.

[25] A.C. Enders, A.O. Welsh, Structural interactions of trophoblast and uterusduring hemochorial placenta formation, J. Exp. Zool. 266 (6) (1993) 578e587.

[26] A. Erlebacher, Immunology of the maternal-fetal interface, Annu. Rev.Immunol. 31 (31) (2013) 387e411.

[27] C. Freyer, U. Zeller, M.B. Renfree, The marsupial placenta: a phylogeneticanalysis, J. Exp. Zool. Part A Comp. Exp. Biol. 299A (1) (2003) 59e77.

[28] S. Furukawa, S. Hayashi, K. Usuda, M. Abe, S. Hagio, I. Ogawa, Toxicologicalpathology in the rat placenta, J. Toxicol. Pathol. 24 (2) (2011) 95e111.

[29] B. Gellersen, J.J. Brosens, Cyclic decidualization of the human endometrium inreproductive health and failure, Endocr. Rev. 35 (6) (2014) 851e905.

[30] D. Gersell, F. Kraus, Diseases of the placenta, in: R. Kurman, L. Ellenson,B. Ronnett (Eds.), Blaustein's Pathology of the Female Genital Tract, Springer,US, 2011, pp. 999e1073.

[31] A. Gopalakrishna, M.S. Khaparde, Development of the foetal membranes andplacentation in the Indian false vampire bat,Megaderma lyra lyra (Geoffroy),Proc. Indian Acad. Sci. - Sect. B Anim. Sci. 87 (9) (1978) 179e194.

[32] O. Grosser, Fruhentwicklung, Eihautbildung und Placentatsion des Menschenund der Saugetiere, J. F. Bermann, Munchen, 1927.

[33] Y. Gu, G.M. Jow, B.C. Moulton, C. Lee, J.A. Sensibar, O.K. Park-Sarge, T.J. Chen,G. Gibori, Apoptosis in decidual tissue regression and reorganization, Endo-crinology 135 (3) (1994) 1272e1279.

[34] D.W. Hone, M.J. Benton, The evolution of large size: how does Cope's rulework? Trends Ecol. Evol. 20 (1) (2005) 4e6.

[35] N.A. Johnson, A.H. Porter, Towards a new synthesis: population genetics andevolutionary developmental biology, Genetica 112e113 (2001) 45e58.

[36] D.B. Keskin, D.S.J. Allan, B. Rybalov, M.M. Andzelm, J.N.H. Stern, H.D. Kopcow,L.A. Koopman, J.L. Strominger, TGF beta promotes conversion of CD16(þ)peripheral blood NK cells into CD16(-) NK cells with similarities to decidualNK cells, Proc. Natl. Acad. Sci. U. S. A. 104 (9) (2007) 3378e3383.

[37] Z. Kielan-Jaworowska, R.L. Cifelli, Z.-X. Luo, Mammals from the Age of Di-nosaurs: Origins, Evolution, and Structure, Columbia University Press, NewYork, 2004.

[38] J.E. Kihlstrom, Period of gestation and body-weight in some placental mam-mals, Comp. Biochem. Physiol. 43 (Na3) (1972) 673.

[39] K. Kin, J. Maziarz, G.P. Wagner, Immunohistological study of the endometrialstromal fibroblasts in the opossum, Monodelphis domestica: evidence forhomology with eutherian stromal fibroblasts, Biol. Reprod. 90 (5) (2014) 111.

[40] K. Kitaya, J. Yasuda, I. Yagi, Y. Tada, S. Fushiki, H. Honjo, IL-15 expression athuman endometrium and decidua, Biol. Reprod. 63 (3) (2000) 683e687.

[41] A.P. Labhsetwar, A.C. Enders, Progesterone in corpus luteum and placenta ofarmadillo dasypus novemcinctus, J. Reprod. Fertil. 16 (3) (1968) 381.

[42] I. Lueders, C. Niemuller, P. Rich, C. Gray, R. Hermes, F. Goeritz, T.B. Hildebrandt,Gestating for 22 months: luteal development and pregnancy maintenance inelephants, Proc. R. Soc. B Biol. Sci. 279 (1743) (2012) 3687e3696.

[43] V.J. Lynch, K. Brayer, B. Gellersen, G.P. Wagner, HoxA-11 and FOXO1A coop-erate to regulate decidual prolactin expression: towards inferring the coretranscriptional regulators of decidual genes, PLoS One 4 (9) (2009) e6845.

[44] V.J. Lynch, G. May, G.P. Wagner, Regulatory evolution through divergence of aphosphoswitch in the transcription factor CEBPB, Nature 480 (7377) (2011)383e386.

[45] N.S. Macklon, J.J. Brosens, The human endometrium as a sensor of embryoquality, Biol. Reprod. 91 (4) (2014) 98.

[46] A. Malassine, R. Leiser, Morphogenesis and fine-structure of the near termplacenta of Talpa Europaea .1. Endotheliochorial labyrinth, Placenta 5 (2)(1984) 145e158.

[47] R. Martin, Evolution of placentation in primates: implications of mammalianphylogeny, Evol. Biol. 35 (2) (2008) 125e145.

A.R. Chavan et al. / Placenta 40 (2016) 40e51 51

[48] R.D. Martin, A.M. Maclarnon, Gestation period, neonatal size and maternalinvestment in placental mammals, Nature 313 (5999) (1985) 220e223.

[49] B.M. McAllan, Chapter 10-reproductive endocrinology of prototherians andmetatherians, in: D.O.N.H. Lopez (Ed.), Hormones and Reproduction of Ver-tebrates, Academic Press, London, 2011, pp. 195e214.

[50] A. Mess, A.M. Carter, Evolutionary transformations of fetal membrane char-acters in Eutheria with special reference to Afrotheria, J. Exp. Zool. B Mol. Dev.Evol. 306 (2) (2006) 140e163.

[51] G. Mor, I. Cardenas, V. Abrahams, S. Guller, Inflammation and pregnancy: therole of the immune system at the implantation site, Ann. N. Y. Acad. Sci. 1221(1) (2011) 80e87.

[52] H.W. Mossman, Comparative morphogenesis of the fetal membranes andaccessory uterine structures, Contrib. Embryol. 26 (158) (1937) 133e137.

[53] H.W. Mossman, Vertebrate Fetal Membranes : Comparative Ontogeny andMorphology, Evolution, Phylogenetic Significance, Basic Functions, ResearchOpportunities, Rutgers University Press, New Brunswick, N.J, 1987.

[54] K. Nakakura, N.M. Czekala, B.L. Lasley, K. Benirschke, Fetal maternal gradientsof steroid-hormones in the 9-banded armadillo (Dasypus-Novemcinctus),J. Reprod. Fertil. 66 (2) (1982) 635e643.

[55] P. Nancy, E. Tagliani, C.S. Tay, P. Asp, D.E. Levy, A. Erlebacher, Chemokine genesilencing in decidual stromal cells limits T cell access to the maternal-fetalinterface, Science 336 (6086) (2012) 1317e1321.

[56] M.J. Novacek, G.W. Rougier, J.R. Wible, M.C. McKenna, D. Dashzeveg,I. Horovitz, Epipubic bones in eutherian mammals from the Late Cretaceous ofMongolia, Nature 389 (6650) (1997) 483e486.

[57] M.A. O'Leary, J.I. Bloch, J.J. Flynn, T.J. Gaudin, A. Giallombardo, N.P. Giannini,S.L. Goldberg, B.P. Kraatz, Z.X. Luo, J. Meng, X. Ni, M.J. Novacek, F.A. Perini,Z.S. Randall, G.W. Rougier, E.J. Sargis, M.T. Silcox, N.B. Simmons, M. Spaulding,P.M. Velazco, M. Weksler, J.R. Wible, A.L. Cirranello, The placental mammalancestor and the post-K-Pg radiation of placentals, Science 339 (6120) (2013)662e667.

[58] M. Pavlicev, G.P. Wagner, A model of developmental evolution: selection,pleiotropy and compensation, Trends Ecol. Evol. 27 (6) (2012) 316e322.

[59] A. Portmann, Die Ontogenese der S€augetiere als Evolutionsproblem, S. Karger,Basel, 1938.

[60] S.M. Reilly, E.J. McElroy, T.D. White, Abdominal muscle function in ventilationand locomotion in new world opossums and basal eutherians: breathing andrunning with and without epipubic bones, J. Morphol. 270 (2009) 1014e1028.

[61] S.M. Reilly, T.D. White, Hypaxial motor patterns and the function of epipubicbones in primitive mammals, Science 299 (5605) (2003) 400e402.

[62] M. Renfree, Endocrinology of pregnancy, parturition and lactation in marsu-pials, in: G.E. Lamming (Ed.), Marshall’s Physiology of Reproduction, Springer,Netherlands, 1994, pp. 677e766.

[63] K.D. Rose, Skeleton of diacodexis, oldest known artiodactyl, Science 216(4546) (1982) 621e623.

[64] K.D. Rose, J.D. Archibald, The Rise of Placental Mammals: Origins and Re-lationships of the Major Extant Clades, JHU Press, 2005.

[65] M.S. Salker, J. Nautiyal, J.H. Steel, Z. Webster, S. Sucurovic, M. Nicou, Y. Singh,

E.S. Lucas, K. Murakami, Y.W. Chan, S. James, Y. Abdallah, M. Christian,B.A. Croy, B. Mulac-Jericevic, S. Quenby, J.J. Brosens, Disordered IL-33/ST2activation in decidualizing stromal cells prolongs uterine receptivity inwomen with recurrent pregnancy loss, PLoS One 7 (12) (2012) e52252.

[66] C. Shooner, P.L. Caron, G. Frechette-Frigon, V. Leblanc, M.C. Dery, E. Asselin,TGF-beta expression during rat pregnancy and activity on decidual cell sur-vival, Reprod. Biol. Endocrinol. 3 (2005) 20.

[67] N.B. Simmons, K.L. Seymour, J. Habersetzer, G.F. Gunnell, Primitive earlyeocene bat from wyoming and the evolution of flight and echolocation, Na-ture 451 (7180) (2008) 818e821.

[68] R.S. Smith, T.J. Smith, T.M. Blieden, R.P. Phipps, Fibroblasts as sentinel cells -synthesis of chemokines and regulation of inflammation, Am. J. Pathol. 151 (2)(1997) 317e322.

[69] F.J. Stansfield, W.R. Allen, Luteal maintenance of pregnancy in the Africanelephant (Loxodonta africana), Reproduction 143 (6) (2012) 845e854.

[70] K.A. Swaggart, M. Pavlicev, L.J. Muglia, Genomics of preterm birth, Cold SpringHarb. Perspect. Med. 5 (2) (2015).

[71] I. Tamura, Y. Ohkawa, T. Sato, M. Suyama, K. Jozaki, M. Okada, L. Lee,R. Maekawa, H. Asada, S. Sato, Y. Yamagata, H. Tamura, N. Sugino, Genome-wide analysis of histone modifications in human endometrial stromal cells,Mol. Endocrinol. 28 (10) (2014) 1656e1669.

[72] D. Thursby-Pelham, The placentation of Hyrax capensis, Philos. Trans. R. Soc.Lond. Ser. B Contain. Pap. a Biol. Character 213 (1925) 1e20.

[73] J.R. True, E.S. Haag, Developmental system drift and flexibility in evolutionarytrajectories, Evol. Dev. 3 (2) (2001) 109e119.

[74] P. Vacca, C. Vitale, E. Montaldo, R. Conte, C. Cantoni, E. Fulcheri, V. Darretta,L. Moretta, M.C. Mingari, CD34þ hematopoietic precursors are present inhuman decidua and differentiate into natural killer cells upon interaction withstromal cells, Proc. Natl. Acad. Sci. U. S. A. 108 (6) (2011) 2402e2407.

[75] R. V�azquez-Molinero, T. Martin, M. Fischer, R. Frey, Comparative anatomicalinvestigations of the postcranial skeleton of Henkelotherium guimarotaeKrebs, 1991 (Eupantotheria, Mammalia) and their implications for its loco-motion, Zoosyst. Evol. 77 (2) (2001) 207e216.

[76] G.P. Wagner, V.J. Lynch, Evolutionary novelties, Curr. Biol. 20 (2) (2010)R48eR52.

[77] H. Wang, S.K. Dey, Roadmap to embryo implantation: clues from mousemodels, Nat. Rev. Genet. 7 (3) (2006) 185e199.

[78] K.M. Weiss, S.M. Fullerton, Phenogenetic drift and the evolution of genoty-peephenotype relationships, Theor. Popul. Biol. 57 (3) (2000) 187e195.

[79] D.E. Wildman, C.Y. Chen, O. Erez, L.I. Grossman, M. Goodman, R. Romero,Evolution of the mammalian placenta revealed by phylogenetic analysis, Proc.Natl. Acad. Sci. U. S. A. 103 (9) (2006) 3203e3208.

[80] W.A. Wimsatt, A.C. Enders, Structure and morphogenesis of the uterus,placenta, and paraplacental organs of the neotropical disc-winged batthyroptera-tricolor-spix (Microchiroptera, Thyropteridae), Am. J. Anat. 159 (2)(1980) 209e243.

[81] F.B.P. Wooding, G. Burton, Comparative Placentation: Structures, Functionsand Evolution, Springer, Berlin, 2008.